A library catalogue is often mistaken for a list.
It is much more useful to think of it as an identity and routing system.
A catalogue tells the library what it has, tells the user what can be found, distinguishes one work from another, connects different versions of the same work, records who created what, and points from an abstract description to a physical shelf, a digital file, an external database or another institution.
This article is part of eduKateSG’s How X Works programme and the How a Library Works series.
The shortest useful answer
A library catalogue works by creating structured records that identify, describe, relate and locate library resources, then indexing those records so users can search and browse them from many different starting points.
That means a catalogue must solve four different problems at once:
- Identity: what exactly is this?
- Description: what do we know about it?
- Relationship: what is it connected to?
- Location or access: how can the user reach it?
Why a title is not enough
Suppose a user asks for a book called Introduction to Mathematics.
That phrase alone may be insufficient. There may be many works with similar titles. The user may need a particular author, edition, language, publisher or year. One edition may contain a new chapter. Another may be a translation. A third may be an ebook licensed only to university users.
The catalogue’s job is to prevent these objects from collapsing into one ambiguous label.
It does this by storing multiple attributes together in a structured record.
The bibliographic record
A bibliographic record is a structured description of a resource. Depending on the resource and the cataloguing standard, the record may include:
- title and subtitle;
- creator or contributors;
- edition statement;
- publication or production details;
- date;
- physical description or file characteristics;
- language;
- series information;
- notes;
- subject terms;
- classification numbers;
- identifiers such as ISBNs or other persistent identifiers;
- links to related works, editions or online access;
- holdings and location information.
The purpose is not to record everything that could possibly be said. It is to record enough structured information to distinguish, connect and retrieve.
Cataloguing creates a stable identity layer
Libraries have a persistent problem: names and titles vary.
An author may publish under initials, a full name, a married name, a pen name or different transliterations. An organisation may change its official name. A classical work may exist under many translated titles. A historical place may have several names across time.
If the catalogue treated every textual variation as a completely unrelated entity, discovery would fragment.
Cataloguing therefore works partly by separating the thing from the strings used to name the thing.
Once that distinction is made, many name variants can point toward one controlled identity.
Authority control: keeping names coherent
Authority control is the mechanism libraries use to make names and subjects consistent enough to retrieve reliably.
An authority record may establish a preferred form of a person’s name and record alternative forms. It may distinguish two people with identical names. It may connect an organisation’s current name with earlier names. It may establish preferred subject terminology and relationships to broader, narrower or related terms.
This matters because search should not require the user to know the catalogue’s exact spelling before the search begins.
Works, editions and copies are not the same thing
One of the catalogue’s hardest jobs is representing sameness at several levels.
A novel can be understood as an intellectual work. That work may appear as a particular edition. The edition may then exist as several physical copies in different branches. A digital version may be available under another licence.
Users move among these levels constantly without naming them.
- “Do you have this book?” may mean the work.
- “Do you have the newest one?” may mean an edition.
- “Is there a copy available at Punggol?” refers to an item or holding.
- “Can I read it online?” refers to a digital manifestation and an access right.
A good catalogue preserves these distinctions while presenting them in a way ordinary users can navigate.
Metadata is the catalogue’s working language
Metadata is often described as data about data. In libraries, a more useful description is structured information that allows an object to participate in a larger system.
Without metadata, a digital file may be only a filename. With metadata, the system can know what the file represents, who created it, when it was created, what rights apply, which collection it belongs to, what subjects it covers and how it relates to other resources.
Metadata makes machine operations possible because important distinctions are placed into predictable fields rather than left buried in prose.
Structured fields change what search can do
Imagine a catalogue record as one long paragraph. Search could still find words, but it would have difficulty distinguishing an author’s name from a publisher’s name, a publication year from a historical date discussed in the book, or a subject heading from a note.
Structured fields let the system ask more precise questions:
- find works by this creator;
- limit to items published after a certain year;
- show only resources in this language;
- filter for ebooks;
- retrieve records assigned to this subject;
- sort editions chronologically;
- show only resources currently available in a branch.
Search quality therefore depends not only on a clever search algorithm but on the structure of the underlying records.
The catalogue is not the collection
This is an important distinction.
The collection consists of books, journals, archives, media, files and other resources. The catalogue consists of representations of those resources.
The record is not the book. It is a model of the book designed for identification and discovery.
Because it is a model, the catalogue can be wrong, incomplete or out of sync. A record may say a book is on the shelf when it is missing. A link may point to an expired resource. An author may have been identified incorrectly. A subject term may be too broad.
Libraries therefore maintain the relationship between record and reality continuously.
How a search moves through a catalogue
When a user enters a query, the visible search can trigger several hidden operations.
- The query is received and normalised.
- The system may correct spelling or expand variants.
- Indexed catalogue fields are searched.
- Matches are scored for relevance.
- Availability, format, date or other facets may be applied.
- Records are grouped or deduplicated where possible.
- The interface presents results with enough context for the user to choose.
- The selected record routes the user to a shelf location, request action, full text, database or external service.
The search result is therefore a route generated from metadata.
Known-item search and exploratory search
Catalogues must support very different user behaviours.
In a known-item search, the user already has a title, author, ISBN or citation. Precision matters. The catalogue needs to identify the exact object quickly.
In exploratory search, the user begins with a topic. Recall and navigation matter. The catalogue should expose subject relationships, related works, classification neighbourhoods and useful filters.
This is why classification and cataloguing complement each other. The record gives the item identity; classification gives it neighbourhood.
The catalogue as a graph
Traditional catalogues are often displayed as records in a list. Underneath, the more powerful model is a graph of entities and relationships.
- A person created a work.
- A work has an edition.
- An edition was published by an organisation.
- A copy is held by a library.
- A work has subject a concept.
- A concept is narrower than another concept.
- A digital object is accessible through a platform.
Once relationships are explicit, discovery improves because the user can traverse the network rather than relying only on keyword matches.
Identifiers reduce ambiguity across systems
Names are readable but unstable. Identifiers help machines preserve identity.
An ISBN can identify a particular book edition. Other identifiers may represent articles, people, organisations, datasets or digital objects. Library systems also assign local record and item identifiers.
The purpose is not to replace names but to give systems a more reliable anchor when names vary.
Identifiers are especially important when records move between libraries, publishers, databases and repositories.
Shared cataloguing: why libraries do not describe everything alone
Millions of libraries hold overlapping material. It would be wasteful for every institution to create every bibliographic description from zero.
Shared cataloguing systems and metadata networks let libraries reuse and adapt records. One institution may create a high-quality description that many others import, while each library adds its own holdings and local information.
This creates enormous economies of coordination.
It also creates dependencies. If shared data contains an error, the error can propagate. Libraries therefore need both interoperability and quality control.
The union catalogue: many libraries, one discovery layer
A union catalogue combines records or holdings from multiple institutions so users can discover resources beyond one local collection.
This changes the question from “does my library own it?” to “where in the network can it be found?”
That is the foundation for services such as interlibrary loan and cooperative resource sharing. The catalogue becomes a routing layer across institutions.
Digital access makes the record executable
In a traditional catalogue, a record points to a physical location. In a digital environment, the record can become an action surface.
The user can click to open full text, authenticate into a database, place a hold, request digitisation, export a citation, save a record, follow a persistent identifier or move into a related collection.
This is a significant change. Metadata is no longer only descriptive. It can drive transactions.
Why catalogues sometimes fail users
Catalogue failure is rarely one thing. Common failure modes include:
- poor or missing metadata;
- duplicate records that should be grouped;
- different works incorrectly merged;
- outdated subject terminology;
- broken links;
- holdings that no longer match reality;
- search ranking that overweights keyword frequency;
- interfaces that expose fields without helping users understand them;
- licensed resources that appear discoverable but are inaccessible to the current user.
The important point is that discovery quality is a property of the whole chain, not merely the search box.
Catalogues and AI
AI can make catalogues easier to query because people no longer need to formulate perfect keywords. A conversational system can interpret intent, generate synonyms, combine concepts, summarise results and propose search routes.
But reliable AI discovery still benefits from catalogue structure.
When names are disambiguated, subjects are controlled, identifiers are stable and relationships are explicit, an AI system can reason over firmer ground. When those structures are absent, the model is forced to infer more from unstructured text and may connect the wrong entities.
The catalogue therefore becomes more, not less, valuable in a machine-mediated knowledge environment. It supplies durable identity beneath flexible language.
The catalogue as institutional memory
A catalogue also remembers the collection’s history.
Records can reveal former editions, withdrawn holdings, provenance notes, donor information, changes in title, earlier publishers and relationships among collections. For archives and rare books, descriptive records may become an essential layer of historical evidence in their own right.
This means cataloguing is not only about present retrieval. It is part of preserving context through time.
The complete mechanism
- A resource enters the library through acquisition or another controlled route.
- The cataloguer determines what kind of resource it is.
- Core descriptive data is recorded in structured fields.
- Creators, subjects and other entities are linked to controlled identities where possible.
- Relationships among works, editions, copies, series and formats are represented.
- Classification assigns the item an intellectual and often physical neighbourhood.
- Holdings and access information connect the record to an actionable location.
- Indexes make selected fields searchable.
- The discovery interface ranks, filters and displays results.
- Users move from record to resource.
- Corrections, new information and collection changes feed back into record maintenance.
That is how a library catalogue works.
It is an engineered bridge between a messy world of books, files, people, editions, subjects and institutions and the much simpler question a user wants answered: can you help me find the right thing?